Training rail accident investigators in UK

Journal of Hazardous Materials 111 (2004) 123–129

Training rail accident investigators in UK�

Stephen Watson∗

Arthur D. Little Limited, Science Park, Milton Road, Cambridge CB4 0XL, UK

Available online 27 April 2004

Abstract

In the UK a number of high severity rail accidents, combined with structural change in the rail industry led to the need for revised accidentinvestigation procedures. The revised procedure addressed identified issues with accident investigation, but to ensure maximum benefit interms of identification of underlying causes of accidents and the production of good recommendations, training for rail industry accidentinvestigators was required.

The development of training followed the development of the revised procedure, with key decisions including the scope of the training andthe delivery method that promotes the highest degree of information transfer and individual learning. The training has been delivered to therail industry with considerable success but has, and continues, to face challenges as the industry structure continues to evolve.© 2004 Elsevier B.V. All rights reserved.

Keywords: Rail accidents; Training; Information transfer

1. Introduction

The rail industry in Great Britain has a long history ofstructural change with both consolidation (in the form ofnationalisation in 1946) and fragmentation. British Rail wasprivatised in the 1994–1996 period and split into over 100companies—responsibility for the infrastructure and trainswas split. The fragmentation of the rail industry was op-posed by the Unions, some politicians and many involvedin the industry, but was pushed through by the politicalparty in power. The railway system in Northern Ireland hasalways been separate from British Rail, and is subject to aseparate legal regime.

As part of the privatisation process, railtrack became theinfrastructure controller for the majority of the national railnetwork—London underground continued to be both infras-tructure controller and operator on the underground systemin London. Within railtrack a separate group with no com-mercial interests called safety and standards directorate wascreated, reporting directly to the chairman. One of the func-tions of this directorate was to develop and publish railwaygroup standards on behalf of the industry. These standardsare mandatory standards that apply to railtrack and all train

� The views expressed in this paper are the author’s views and donot necessarily reflect the views of Arthur D. Little or Rail Safety andStandards Board.

∗ Tel.: +44-1223-392271; fax:+44-1223-420021.E-mail address: [email protected] (S. Watson).

and station operators and were designed to help managerisks that could be passed between infrastructure controller,train and station operators. One of these standards definedthe process for industry accident investigation.

The health and safety executive (HSE) is the safety regu-lator for all workplaces in Great Britain and it has responsi-bility for ensuring implementation of the health and safety atwork etc. Act 1974. HM railway inspectorate (part of HSE)is responsible for safety regulation of the rail industry. It isresponsible for giving approval for train companies (infras-tructure operators, train operators and station operators) tooperate (called railway safety cases) and investigates rail-way accidents and can bring prosecutions under the healthand safety at work etc. Act 1974 for alleged breaches ofsafety legislation. The British transport police are a policeforce within Great Britain dedicated to the railways, andthey have a statutory duty to investigate criminal acts on therailway, including allegations of murder, manslaughter andgross negligence, under criminal law.

Since privatisation the rail industry has been under enor-mous public scrutiny in the media and in Parliament, withvery few outside the industry understanding the new struc-tures and roles of different companies.

2. Accident investigation within the rail industry

The rail industry has always had its own accident investi-gation process that has sat alongside the processes operated

0304-3894/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.jhazmat.2004.02.026

124 S. Watson / Journal of Hazardous Materials 111 (2004) 123–129

by HM railway inspectorate. As part of the privatisationprocess a railway group standard was developed to coverthe accident investigation and inquiry process. These indus-try processes were strained by the increasing tendency ofthe railway inspectorate to seek prosecution following acci-dents and the injection of commercial interests as a result ofprivatisation. The Southall accident brought these strains,particularly the tendency to seek prosecution, to a head.

3. Southall accident

At 13:15 on 19 September 1997 a side-on collision oc-curred between the 10:32 Swansea to Paddington high speedtrain, operated by Great Western trains and a freight trainoperated by English Welsh and Scottish, at Southall EastJunction, West London. The driver of the high speed trainhad passed signal SN254 at danger (SPAD).

The collision caused extensive damage to the power carand a number of the coaches of the high speed train, exten-sive damage to the trailing freight wagons with further dam-age to the track and overhead line equipment. Seven peoplewere killed and 139 persons were injured in varying degreesof severity.

Professor John Uff QC was asked by the chair of thehealth and safety commission to lead a public inquiry todetermine: why the accident happened, and in particular toascertain the cause or causes; to identify any lessons whichhave relevance for those with responsibilities for securingrailway safety; and to make recommendations.

The driver of the high speed train was arrested by theBritish transport police on suspicion of manslaughter on 19September 1997 and was released on bail. The public inquiryproceedings began in December 1997 with a formal openingin February 1998.

The driver of the high speed train was charged with sevencounts of manslaughter on 17 April 1998, but no furtherprogress could be made by the public inquiry pending de-cisions on criminal charges being considered against GreatWestern Trains. On 1 December 1998 Great Western Trainswere charged by the health and safety executive with ‘cor-porate manslaughter’ and with offences under the health andsafety at work Act etc.1974.

The trial of the manslaughter charges began at the cen-tral criminal court (Old Bailey) on 21 June 1999 and 6days later the trial Judge rejected the prosecution’ corporatemanslaughter case. On 2 July 1999 Great Western Trainspleaded ‘guilty’ to the charges under the health and safety atwork etc. Act and the Crown and HSE decided to abandonits action against the driver of the high speed train. GreatWestern Trains were fined £1.5 million and were ordered topay the prosecution costs.

The public inquiry commenced hearings on 20 Septem-ber 1999 (just over 2 years after the accident), and wereadjourned on 30 September for a week at the request of thepassenger groups who required more time. On 5 October

1999 a major rail collision occurred at Ladbroke Grove.Hearings of the Southall public inquiry resumed on 25October and were concluded on 25 November with finalsubmission on 20 December 1999.

Professor Uff QC made 93 recommendations under 12separate headings in ‘the Southall rail accident inquiryreport’ published by the health and safety commission in2000. The accident investigation recommendations focusedon the need for the rail industry inquiry to proceed withoutdelay and laid the building blocks for Lord Cullen to rec-ommend creation of an independent accident investigationbody in the Ladbroke Grove part 2 report.

3.1. The need for change

The need for improvements in process of rail accidentinvestigation were driven by five key factors.

• Increased tendency by safety regulator to seek prosecu-tion.

• Results of fragmentation and privatisation of UK rail in-dustry.

• Lack of opportunities for learning about accident investi-gation.

• The delays in the Southall accident investigation due tothe legal process.

• Increased public scrutiny of rail industry.

The industry response to these drivers, led by railtracksafety and standards directorate (which became railwaysafety and now the rail safety standards board) was to sub-stantially revise the railway group standard defining theprocess, and to instigate training in the industry accidentinvestigation process and accident investigation techniques.

4. The new railway group standard

The railway group standard was fully revised and afterseveral rounds of vigorous industry consultation and com-ment the new railway group standard was formally publishedin December 2001, and came into force in February 2002.

During the process of revising the Standard the Lad-broke Grove accident occurred (see later) and the revisedrailway group standard addressed the key accident inves-tigation recommendations for the industry. The processof revising the standard went in parallel to the LadbrokeGrove public inquiry (which reported some 2 years after theaccident).

The overall purpose of the railway group standard is to:

• provide a consistent, comprehensive and structured pro-cess for the investigation of accidents/incidents in orderto prevent, or reduce the risk of their recurrence, withoutapportioning blame or liability;

• provide for the independent investigation of the most se-rious accidents/incidents;

S. Watson / Journal of Hazardous Materials 111 (2004) 123–129 125

Fig. 1. Investigation hierarchy.

• enable information obtained from inquiries/investigationsto be shared with, and used by, organisations with a di-rect responsibility for maintaining, or improving, railwaysafety.

The standard also tied down a number of definitions andtypes of inquiry/investigation. Three types of investigationsare defined:

• Formal inquiry.• Formal investigation.• Local investigation.

The standard defines for these types of investigation theprocesses to be followed, the leadership and composition ofinvestigation teams and the reporting requirements. The for-mal inquiry and formal investigation types have been furthersub-divided in actual practice and this leads to a hierarchyof investigations (Fig. 1).

To illustrate these types with typical accidents:

Formal inquiryFully independent panel Multi-fatality passenger

accidentIndependent chairperson Track worker fatality/

serious injury

Formal investigationProfessional investigator led High-potential signal

passed at dangerIndustry team Operating irregularity

Local investigation On-depot incident

The standard provides some flexibility in determining thetype of investigation, usually based on the learning potential.

The railway ethos of trying to find out what happened andlearn the lessons has been captured in the process definition.One independent chairperson of formal inquiries describedthe sentiments underpinning the operation of the process thatin his experience all the industries parties has always been‘prepared to put commercial interests to one side’.

For all formal inquiries and formal investigations thereare four principal objectives:

• to establish the facts;• to determine the immediate causes of the accident;• to determine the underlying causes of the accident;• to develop robust recommendations.

The standard also took the opportunity to fix a numberof terms to increase consistency—two key terms are theimmediate and underlying causes.

4.1. The immediate cause(s)

“The immediate cause(s) is an unsafe act or unsafe con-dition which causes an accident or incident”

4.2. The underlying cause(s)

“Underlying cause(s) are any factors which led to theimmediate causes of accidents or incidents, or resulted insuch causes not being identified and mitigated”

The revision of the process (as defined by the standard)and development and early deliveries of the training hap-pened in parallel, and then as the standard was finally pub-lished the training courses were able to immediately reflectthe new standard.

5. Training

The tender for development and delivery of training in railaccident investigation was sent out by railtrack safety andstandards directorate in March 1999 and attracted a largenumber of respondents. After an evaluation process the con-tract was awarded to Arthur D. Little in November 1999. Oneof the key factors in awarding this contract was that ArthurD. Little was not involved in rail accident investigation, but


has significant rail industry experience and experience ofaccident investigation in the process and energy industries.

The training was called accident investigation/formal in-quiry training and was aimed at all involved in rail accidentinvestigation within the industry (investigation leaders andteam members, safety directors and those responsible forassessing and acting upon investigation recommendations).There are other training providers covering rail accident in-vestigation and the practical aspects of evidence collection,but this accident investigation/formal inquiry training is usedby rail safety and standards board (formerly railway safety)to help develop incoming formal inquiry chairpersons andis regularly reviewed by it to ensure it meets current needs.

Just before the award of the contract for the training, theLadbroke Grove accident occurred, and the public inquirywas held as the training was being launched.

5.1. Ladbroke Grove

At 08:09 on 5 October 1999 a head-on collision occurredbetween 08:06 Paddington to Bedwyn Thames train turboand the 06:03 Cheltenham to Paddington First Great Westernhigh speed train at Ladbroke Grove junction. The combinedspeed of impact was about 130 mph, and the driver of theThames turbo had passed signal SN109 at danger. Immedi-ately following the crash, diesel from the high speed trainignited causing a severe fire. Thirty-one people died as a re-sult of the crash (24 on the Turbo and 7 on the High SpeedTrain) and a further 227 people were admitted to hospitalsfollowing injury.

Lord Cullen, who had chaired the piper alpha public in-quiry, was appointed by the health and safety commissionto conduct a public inquiry. Lord Cullen split the inquiryinto two main parts, with part 1 looking at the actual ac-cident and part 2 taking a much wider look at the industrypost-privatisation.

Lord Cullen’s part 2 report was published in 2001 (some2 years after the actual accident) and it explores a num-ber of key organisational and regulatory themes for thepost-privatisation rail industry and covered accident inves-tigation. Phrases in the report about accident investigationinclude:

• ‘It is inappropriate for the safety regulator to carry out thefunction of investigation since it might be necessary forthe investigator to examine the decisions and activities ofthe safety regulator itself’.

• ‘It is clear that in general the overriding public interest liesin the swift determination of the causes of rail accidents,the publication of the report and the implementation ofonly safety lessons’.

• ‘Applying and disseminating the lessons of accidents andincidents (including near misses). . . was inhibited by the‘blame culture’, and lack of a co-ordinated system forthe collation of recommendations and ensuring they werefollowed up’.

A number of the 74 recommendations in the part2 report related to accident investigation, and two areparticularly related to the existing rail industry pro-cesses.

62 The sole objective of the investigation of accidentsor incidents should be the prevention of accidentsand incidents. It should not be the purpose of suchinvestigations to apportion blame or liability.

63 The appointment of an independent chairman and,where appropriate, independent members for thepanel of a formal inquiry, is endorsed.

In addition to recommendations on the rail industry’s ownprocesses, Lord Cullen took the suggestion from ProfessorUff’s Southall recommendations on the need for an inde-pendent investigation body and made nine key recommen-dation about the creation of a rail accident investigationbody.

57 The responsibility of the health and safetyexecutive (HSE) for the investigation of railaccidents should be transferred to an independentbody, here referred to for convenience as therailway accident investigation branch (RAIB).

58 The investigation of rail accidents and incidents ofwhatever nature should be brought under theoverall control of the RAIB.

59 The more serious cases should be the subject ofinquiry by the RAIB. The categories of case whichwould fall to the RAIB to inquire into should be thesubject of further study.

60 The less serious cases should be delegated to theindustry to be dealt with by formal inquiry orformal investigation. However, the RAIB shouldhave the ability to call in any case for inquiry byitself where that appears to be appropriate.

67 The RAIB should exercise a supervisory functionin regard to the working of formal inquiries andformal investigations.

69 The reports of RAIB inquiries and formal inquiriesshould be published, subject to the protection of theidentity of persons involved.

70 The rail industry safety body should maintain acurrent record of:

a. the recommendations of RAIB inquiries andformal inquiries;b. the responses of all the organisations to whichthe respective recommendations are directed; andc. the state of progress towards implementationin relation to stated timescales.

71 The RAIB should regularly examine the reports offormal investigations in order to determine whetherthere are matters of importance which should bebrought to the attention of the industry.


73 The statements made by witnesses in connectionwith RAIB inquiries and industry inquiries andinvestigations should not be disclosed to the police,save by order of a judge.

The creation of the rail accident investigation branch re-quired changes to the existing legislation and these propos-als are now going through the Parliamentary process. Theimpact of the RAIB will be discussed later.

6. Accident investigation/formal inquiry training

The training was designed to cover all the members andleaders of formal investigation panels and formal inquiryteams. An early decision was taken to spilt the training intotwo separate courses.

• A 3-day core course for all members and leaders.• And a 2-day supplementary course for leaders.

The supplementary course would only be accessible tothose that have successfully completed the core course, andinitial estimates suggested that only one in three core coursedelegates would need to attend the supplementary course.

In overview:

Core courseExplanation and practice in applying industry process.Introduction to range of generic accident investigationtools.Evidence collection, interviewing, analysis andconclusion/recommendation.Importance of effective reporting to be understood.Assessment of understanding tested through writtenexam.Guest speaker presentation by an independentchairperson.

Supplementary courseComplexities of technical evidence to be demonstrated.Skills for preparing and running accident investigations.Range of different cases to be shown.Writing effectively to be practised.

6.1. Core course

The core course was designed to provide training in theapplication of the industry process, but it was not designedas a briefing on the process itself. A key decision for thecore course was which investigation techniques should beincluded—there are many commercially available tech-niques, books and software programmes? The railway groupstandard does not provide any guidance on types of tech-niques to be used, as it specifies the process and the desiredoutcomes. After discussion with existing railway accidentinvestigators the decision was to cover four general causal

analysis techniques and then allow the delegates to decidewhich they would use.

The four techniques selected are:

• Analysis of events.• Barrier/defence identification and analysis.• ‘Checklist’ analysis.• Unstructured analysis.

And these four techniques are demonstrated in the corecourse. Central to all Arthur D. Little training courses is theprinciple of using a mixture of briefings and case studies topromote learning, supported by suitable materials. For thecore course, it was decided to use a single case study overthe 3 days, broken into a number of separate elements.

Working with railtrack safety and standards directorate(which became railway safety and in 2003 rail safety andstandards board), a real accident where a track workerwas critically injured when struck by a moving train wasselected. Significant discussions with the actual formalinquiry chairman were held to understand how the acci-dent was investigated and to develop suitable materialsfor the course. About 10% of additional fictional materialwas added to cover several aspects not seen in the actualinvestigation.

The case study is split into six parts:

• Introduction—two page ‘fax’ outlining the incident typeand bare facts.

• Initial evidence gathering—15 pages of evidence thatcould be readily gathered on site.

• Review of on-site interview and safety critical communi-cation—extracts from the on-site interview with the traindriver involved and the transcript of the emergency callfrom the accident site to the remote control centre.

• Role-play interview—opportunity to interview the per-son responsible for the on-site safety of the injured trackworker.

• Factors for consideration—15 pages of evidence that werecollected at the formal inquiry panel meeting.

• Conclusions and recommendations—producing the keyelements of the accident report.

A typical course delivery with 16 delegates will be splitinto four case study groups for the entire 3 days. The casestudy groups are generated by the trainers, to ensure as faras possible a good mix of experience and industry sectorsin each group. For this reason no single company courseshave been run—mixed courses are necessary to reflect thecomposition of actual investigation teams. Supporting thisis a maximum number of delegates from any one companyper course.

Unlike many other Arthur D. Little courses there are nomodel solutions for the case study and there is very lit-tle interference by trainers in the case study work. Thereare often tensions and disagreements within case studyteams, and this is a reflection of reality! At the end of thecourse all the delegates get copies of the conclusions and


recommendations produced by all the groups on the course.This serves to highlight the similarities and differences thatcan evolve from the evidence gathered!

To provide the delegates with further insights into actualinvestigations two sessions beyond the briefings and casestudy are included. A short session at the end of the firstday, in the form of a slide show, illustrates some of thepotential physical and technical challenges of investigatingreal railway accidents. A second session at the end of thesecond day is included to provide an opportunity for a guestspeaker to share their thoughts and answer questions fromthe delegates. For the first year a number of retired seniorofficers from the British transport police were used, but afterdelegate feedback it was decided instead to use actual formalinquiry independent chairmen.

The guest speaker session usually takes the format of a dis-cussion about ‘interesting investigations’ the chairman hasbeen involved in, followed by an open question-and-answersession on railway accident investigation.

This has proved to be a real success, allowing delegatesto ask about the real-life implementation of the process, aswell as providing the chairmen an opportunity to meet in-dustry members who may be involved in future investiga-tions. Railway Safety and Standards Board also sends oneof their senior managers involved in managing the investiga-tion process to demonstrate their support to the training andto listen to the types of issues emerging from the delegates.

The course is residential—we have found that discussionsin the coffee breaks and meals are just as important for com-petence development as the formal sessions. To maximisethe case study time a pre-reading handbook is sent to alldelegates 2 weeks before course commencement.

To conclude the core course, a 90 min closed-book writ-ten exam is used to test delegates understanding of the in-vestigation process. Fifteen questions cover the entire rangeof material used during the course, and the pass mark is setat 60%. It is recognised that this is not a perfect measureof investigation competence, but provides a measure of theunderstanding of the underlying principles and the industryprocess.

6.2. Supplementary course

The supplementary course was designed to introduce thepotential complexities with technical evidence, and to de-velop key skills in leading investigations and writing re-ports. The supplementary course uses three different casestudies illustrating different technical and operational as-pects of the railway. Attendance on core course and passingthe core course exam is a mandatory requirement before at-tendance on the supplementary course. However, unlike thecore course, no exam is used. Testing the competence of in-vestigation leaders with a written exam was not seen to beworthwhile.

The first case study is designed to provide experience ofsetting up an inquiry. Information is provided about a rail-

way accident involving a broken axle on a freight wagon andsubsequent derailment and secondary collision. Delegatesare given details of the organisations involved and writtenreports from direct and indirect witnesses. From this infor-mation delegates are asked to select the organisations to pro-vide inquiry panel members and observes. delegates then goon to decide which staff they would like to interview and todevelop an interview schedule. A subsequent part providesthe delegates with a technical report commissioned from anexternal expert and ask the delegates to review the reportand to decide what to do with it (for example conduct a fur-ther investigation, commission additional studies, or writerecommendations).

The second case study involves a derailment of anover-speeding freight train on a poor section of track. Initialevidence gathered suggests five possible immediate causesand the delegates are provided with a large amount of tech-nical evidence from the various organisations. Delegateshave to review the evidence and decide which possible im-mediate causes can be eliminated and how they would goabout determining the most probable immediate cause. Thecase in unusual as the actual independent chairperson com-missioned two different companies to model the derailmentwith widely different results!

The third case study provides delegates with an op-portunity to practice writing plain English. While it mayseem quite simple, it is in reality very difficult to explaincomplicated operational and technical issues, without ap-portioning blame and in a way that can be understoodby all who might need to read and act upon the accidentreport.

Both the core and supplementary courses are delivered bytwo tutors—one from Arthur D. Little and one from industry.The Arthur D. Little tutor provides strong process knowl-edge, experience of accident investigation, training coursedelivery and facilitation skills for small and large groups,underpinned by a working knowledge of the rail industry.The industry tutor provides many years of practical rail ac-cident investigation experience.

7. The current situation

7.1. The courses

Some 3 years after inception, the core course remainssubstantially unchanged. The materials are revised regularly(now using version 10) but the overall structure and the casestudy have stood the test of time well. Over 250 delegateshave attended 17 courses and feedback shows the course ismeeting delegate needs.

Ongoing delegate feedback, and analysis of the completedexam papers helps to identify areas for improvement.

However, the supplementary course has proved moredifficult in knowing what to include and how to pitch thedelivery. To date seven courses have been run with 90


delegates. Delegate feedback led to it being completelyredesigned in 2002 to reflect changing needs of leaders ofaccident investigation teams—more focus on the potentialtechnical challenges, practice in writing recommendationsand more opportunities to share experiences. The changesmade a step-change in delegate feedback.

The courses however remain a challenge to deliver effec-tively due to the wide number of industry parties and thecontinuing introduction of new people into the industry andaccident investigation positions.

7.2. The future?

The future of the training is uncertain with the creationof the Rail Accident Investigation Branch currently goingthrough the legislative process. It is believed that the exist-ing rail industry processes will continue (this was endorsedby Lord Cullen in the part 2 inquiry into the Ladbroke Groveaccident), and if there is an industry process there will con-tinue to be a need for training in effective application of theprocess.

Documents

Training rail accident investigators in UK